Files
VoiceStudio/tests/test_models_catalog.py
T

144 lines
5.9 KiB
Python

"""Regression guard for the bundled model catalog (``backend/config/models.yaml``).
Issue #239: two ASR entries pointed at Hugging Face repo IDs that returned
HTTP 404 ("Repository Not Found"), so users could not install those ASR models:
* ``UsefulSensors/moonshine-small`` — no such model (Moonshine ships *tiny* /
*base*, there is no 300M "small").
* ``Systran/faster-whisper-large-v3-turbo`` — Systran publishes no turbo repo.
These tests are intentionally **static** (no network) so they run in CI: they
assert every ``repo_id`` is well-formed and that the known-bad IDs can never be
reintroduced. They do not verify live HF availability — that would be flaky and
slow — but they stop the catalog from shipping the exact IDs that broke #239.
"""
import re
from pathlib import Path
import yaml
_YAML = Path(__file__).resolve().parents[1] / "backend" / "config" / "models.yaml"
# org-or-user / repo-name, both segments HF-legal (letters, digits, _, -, .).
_REPO_RE = re.compile(r"^[A-Za-z0-9][\w.-]*/[\w.-]+$")
# Repo IDs that returned HTTP 404 on Hugging Face (issue #239). Must never reappear.
_KNOWN_BAD = {
"UsefulSensors/moonshine-small",
"Systran/faster-whisper-large-v3-turbo",
}
def _models():
data = yaml.safe_load(_YAML.read_text(encoding="utf-8"))
return data["models"] if isinstance(data, dict) and "models" in data else data
def test_catalog_loads_and_has_entries():
models = _models()
assert isinstance(models, list) and len(models) > 0
def test_every_repo_id_is_well_formed():
for m in _models():
rid = m.get("repo_id")
assert rid and _REPO_RE.match(rid), f"malformed repo_id: {rid!r}"
def test_config_only_entries_declare_the_files_that_complete_them():
"""A ``config_only`` repo carries no weights of its own, so the install
validator judges it by ``config_required_files`` instead of a weight floor.
Declare none and the entry is permanently uninstallable: the completeness
check returns False and the install fails with a message whose list of
required files is empty. That is the shape #2163 reported, so it is a
catalogue invariant rather than something a user should discover.
"""
for m in _models():
if not m.get("config_only"):
continue
required = m.get("config_required_files")
assert isinstance(required, list) and required, f"{m['repo_id']}: config_only needs a nonempty config_required_files list"
assert all(
isinstance(name, str) and name.strip() for name in required
), f"{m['repo_id']}: blank entry in config_required_files"
def test_dependency_declarations_are_installable():
"""Every declared dependency needs an id and the files that prove it landed
— the installer rejects a dependency snapshot that lacks them."""
for m in _models():
for dependency in m.get("dependencies") or ():
rid = dependency.get("repo_id")
assert rid and _REPO_RE.match(rid), f"malformed dependency repo_id: {rid!r}"
required = dependency.get("required_files")
assert isinstance(required, list) and required and all(
isinstance(name, str) and name.strip() for name in required
), (
f"{m['repo_id']}{rid}: dependency needs required_files"
)
def test_required_fields_present():
for m in _models():
for field in ("repo_id", "label", "role"):
assert m.get(field), f"{m.get('repo_id')!r} missing required field {field!r}"
def test_known_404_repo_ids_absent():
ids = {m.get("repo_id", "") for m in _models()}
leaked = ids & _KNOWN_BAD
assert not leaked, f"known-404 repo IDs reintroduced (issue #239): {leaked}"
# ── TTS-only required gate + curated_on (per-platform curation) ─────────────
# The app must boot with just the TTS model: ASR is optional and surfaced as
# per-platform curated picks (curated_on) instead of a required download.
_CURATED_TAGS = {"all", "darwin-arm64", "darwin-x86_64", "cuda", "rocm", "cpu"}
def test_only_tts_is_required():
required = [m for m in _models() if m.get("required")]
assert [m["repo_id"] for m in required] == ["k2-fsa/OmniVoice"], (
"exactly one required model is allowed — the TTS engine. ASR models "
f"are optional curated picks (curated_on). Got: {required}"
)
assert required[0]["role"] == "TTS"
def test_curated_on_tags_are_valid():
for m in _models():
for tag in m.get("curated_on", []) or []:
assert tag in _CURATED_TAGS, (
f"{m['repo_id']}: unknown curated_on tag {tag!r} "
f"(valid: {sorted(_CURATED_TAGS)})"
)
def test_every_platform_has_a_curated_asr_pick():
"""Each host family must resolve at least one curated offline-capable ASR
model, or the ASR-missing download CTA would have nothing to offer."""
models = _models()
for family_tags in (
{"darwin", "darwin-arm64"},
{"darwin", "darwin-x86_64", "cpu"},
{"win32", "win32-AMD64", "cuda"},
{"linux", "linux-x86_64", "cuda", "rocm"},
{"linux", "linux-x86_64", "cpu"},
):
curated_asr = [
m for m in models
if m.get("role") == "ASR"
and (
"all" in (m.get("curated_on") or [])
or set(m.get("curated_on") or []) & family_tags
)
and (not m.get("platforms") or set(m["platforms"]) & family_tags)
]
assert curated_asr, f"no curated ASR pick resolves for host tags {family_tags}"
def test_single_cross_platform_dictation_recommendation_is_whisper_tiny():
"""Installer curation must match the runtime/UI dictation default."""
sherpa = [m for m in _models() if m.get("engine") == "sherpa-onnx"]
recommended = [m["dictation_id"] for m in sherpa if "all" in (m.get("curated_on") or [])]
assert recommended == ["sherpa-whisper-tiny"]